Independent R&D Demo · Synthetic Policy Cases · Agency-Neutral

VERITASPolicy-Grounded AI Decision Assurance

VERITAS explores how AI-enabled decisions can be checked against policies, obligations, prohibitions, exceptions, evidence requirements, precedence, and knowledge limits before an action proceeds, while preserving an auditable decision trace.

Policy GroundingObligations & Prohibitions Exception HandlingConflict & Precedence Evidence-Gated DecisionsAuditability
93.8%

held-out synthetic policy-decision accuracy

Evaluated on 400 disjoint held-out cases after a separate 400-case confidence-calibration split. The shallow policy-matching baseline reached 53.8%.

95.2%violation-detection F1 vs. 63.7% baseline
5.0%false-block rate vs. 28.0% baseline
95.8%complete audit traces vs. 41.3% baseline
1.4%held-out confidence calibration error
Interactive decision-assurance walkthrough

Watch a recommendation become an auditable decision.

Choose a synthetic policy scenario or let the Auto Tour cycle through them. The left side shows what a shallow policy-matching approach might conclude. The right side shows the public VERITAS workflow: ground relevant rules, identify normative constraints, check evidence and exceptions, resolve conflict or precedence, then produce an explicit decision and trace.

● Live synthetic scenarioCompliant ActionNo external policy data
AI-proposed action
Release an internal analytics summary to an approved team.

The recommendation must be checked against applicable policy obligations and available evidence before execution.

Available case evidence
01 · Ground
Relevant policyIdentify rules governing the proposed action.
02 · Formalize
Normative constraintsSeparate obligations, prohibitions, permissions, and exceptions.
03 · Verify
Evidence & conditionsCheck whether required evidence and exception conditions are satisfied.
04 · Resolve
Conflict & precedenceApply version, exception, and conflict-handling rules.
05 · Assure
Decision & traceReturn the decision, reasons, confidence, alternatives, and audit evidence.
SHALLOW POLICY-MATCHING BASELINEtransparent comparator
ALLOW

Recommendation appears consistent with policy.

Matched authorization language and found no direct prohibition.

Baseline trace
Matched policyP-12
Evidencerole authorization
Exception checknot evaluated
Conflict checknot evaluated
Confidence86%
Audit completeness42%
VERITAS PUBLIC DEMONSTRATION METHODsimplified nonproprietary workflow
ALLOW

Action satisfies the applicable policy constraints.

Required authorization is verified and the decision can be logged with sufficient evidence.

Decision assurance record
Grounded rulesP-12 · P-31
Normative resultobligations satisfied
Evidence statesufficient
Exception / conflictnone applicable
Alternativenot required
Confidence96%
Trace IDVT-1001
Audit completeness98%
Held-out deterministic synthetic evaluation

Evaluate the hard cases: exceptions, conflicts, missing evidence, and policy precedence.

The benchmark contains 800 deterministic synthetic policy-decision cases. Four hundred cases are used only to calibrate confidence. A disjoint 400-case set is used for every headline result below. Cases span eight classes and explicitly include situations that shallow keyword or rule matching often mishandles.

93.8%

overall decision accuracy

Correct ALLOW, BLOCK, or ESCALATE outcome on held-out cases.

53.8% baseline → 93.8%
95.2%

violation-detection F1

Precision and recall for cases whose correct outcome is BLOCK.

63.7% baseline → 95.2%
94.0%

exception-handling accuracy

Valid and invalid exception scenarios evaluated correctly.

51.0% baseline → 94.0%
91.0%

conflict / knowledge-limit escalation

Cases requiring human review or explicit escalation are recognized.

16.0% baseline → 91.0%
95.8%

complete decision traces

Required synthetic policy, evidence, decision, and reasoning fields present.

41.3% baseline → 95.8%
Held-out benchmark detail

Same 400 cases, transparent baseline vs. public VERITAS method.

MetricShallow baselineVERITAS demo
Overall decision accuracy53.8%93.8%
Violation precision62.1%95.5%
Violation recall65.5%95.0%
Violation F163.7%95.2%
False-block rate on allowable actions28.0%5.0%
Exception-handling accuracy51.0%94.0%
Escalation accuracy16.0%91.0%
Audit-trace completeness41.3%95.8%
Cause / evidence alignment53.7%97.9%
Expected Calibration Error5.6%1.4%
Eight synthetic case classes

Policy logic beyond direct prohibitions

  • Compliant action: all obligations and evidence requirements satisfied.
  • Direct prohibition: an applicable MUST NOT rule blocks the action.
  • Missing prerequisite: required authorization or evidence is absent.
  • Valid exception: an exception overrides the general prohibition because all exception conditions are satisfied.
  • Invalid exception: the exception is invoked but required conditions or evidence are missing.
  • Rule conflict: applicable policies point to incompatible outcomes and require escalation.
  • Policy precedence: a newer or higher-priority rule changes the applicable outcome.
  • Insufficient evidence: the system should not force ALLOW or BLOCK when knowledge limits are reached.
800synthetic policy-decision cases
400confidence-calibration cases
400held-out evaluation cases
8case classes
3decision outcomes: allow, block, escalate
Fixed seedsdeterministic generation
High-level public architecture

Policy to constraint to evidence to decision to trace.

The public architecture shows functional responsibilities only. It intentionally does not disclose proprietary semantic-grounding methods, normative compilation, conflict-resolution logic, optimization procedures, model configurations, or implementation-specific scoring.

01 · Ingest

Decision context

Receive the proposed action, relevant facts, available evidence, actors, resources, and operating context.

02 · Ground

Policy relevance

Identify potentially governing provisions and connect policy concepts to the decision context.

03 · Formalize

Normative constraints

Represent obligations, prohibitions, permissions, exceptions, evidence conditions, and applicable versions.

04 · Evaluate

Candidate decisions

Test the proposed action and feasible alternatives against the grounded constraints and evidence state.

05 · Resolve

Conflict & knowledge limits

Handle exceptions, precedence, incompatible rules, missing evidence, and cases requiring human review.

06 · Assure

Decision evidence

Return the decision, grounded rules, supporting evidence, reasons, confidence, alternatives, and auditable trace.

Research questions

The demo is designed around measurable decision-assurance questions.

1

Did the system ground the right policy?

Measure whether the governing provisions for a synthetic decision are identified and carried into the decision trace.

2

Are violations caught without excessive blocking?

Track violation precision/recall together with the false-block rate on otherwise allowable actions.

3

Are exceptions handled correctly?

Distinguish valid exceptions from incomplete or unsupported attempts to invoke an exception.

4

Does the system recognize knowledge limits?

Escalate conflicts and insufficient-evidence cases rather than fabricating a confident binary outcome.

5

Is the decision auditable?

Require matched policy, evidence state, decision, reason, confidence, and trace metadata for review.

6

Does confidence correspond to correctness?

Calibrate confidence on one split and measure Expected Calibration Error only on disjoint held-out cases.

Sanitized public showcase

Show the evidence. Protect the decision-assurance recipe.

Public on this page

What evaluators can inspect

  • Problem definition and decision-assurance workflow.
  • Generic policy concepts: obligations, prohibitions, permissions, exceptions, precedence, and evidence requirements.
  • Synthetic benchmark structure, held-out split, metrics, and limitations.
  • Illustrative decision records and audit fields.
  • Transparent shallow baseline used for comparison.
Intentionally not disclosed

Protected implementation details

  • Proprietary semantic-grounding or policy-retrieval methods.
  • Internal normative-logic compilation and representation.
  • Detailed conflict-resolution and candidate-selection procedures.
  • Private prompts, model configurations, thresholds, or scoring functions.
  • Source code, private datasets, customer policy documents, or protected integrations.
Multi-domain relevance

Policy-grounded decision assurance is broader than any single industry.

AI Governance & Risk

Check AI-enabled recommendations against organizational policies, evidence requirements, and defined escalation boundaries.

Compliance Operations

Connect operational decisions to explicit rules, exceptions, approvals, and reviewable evidence without relying on opaque reasoning alone.

Financial & Enterprise Controls

Support approval, access, transaction, and model-governance decisions with traceable policy checks and evidence gating.

Healthcare Administration

Apply policy-grounded checks to administrative workflows where permissions, prerequisites, documentation, and human review matter.

Industrial & Safety-Critical Workflows

Make constraints, required evidence, exceptions, and escalation conditions visible before an automated recommendation becomes an action.

Human-AI Decision Systems

Provide reviewers with grounded reasons, alternatives, uncertainty, and audit evidence instead of a recommendation alone.

Research leadership

Systems, formalized decision assurance, and AI/ML in one research team.

SD

Dr. Sajib Datta

Principal Investigator · Systems & Decision-Assurance Lead

Technical direction, policy-to-system architecture, evidence and audit pipelines, experimental design, reproducibility, performance evaluation, commercialization-oriented prototyping, and end-to-end integration.

TD

Dr. Tonmoay Deb

Co-Investigator · AI/ML & Reasoning Lead

Semantic reasoning, trustworthy AI, confidence-aware decision methods, adversarial and edge-case analysis, model evaluation, explanation design, and AI/ML experimentation.

Public technical context

Grounded in explainability, knowledge limits, risk measurement, and documented decision processes.

NIST · AI Risk Management Framework

The AI RMF organizes risk-management activity around Govern, Map, Measure, and Manage and emphasizes continuous risk management across the AI lifecycle.

Public reference ↗
NIST · Four Principles of Explainable AI

NIST identifies explanation, meaningfulness, explanation accuracy, and knowledge limits as fundamental principles for explainable AI systems.

Public reference ↗
NIST · Generative AI Profile

NIST's cross-sector GenAI profile extends AI RMF guidance for managing risks associated with generative AI systems across the lifecycle.

Public reference ↗
Independent, nonproprietary research disclaimer.
VERITAS on this page is an independent Omniscient Innovations LLC research demonstration using deterministic synthetic policies, evidence records, and decision cases. It is agency-neutral and is not sponsored, funded, endorsed, certified, selected, or validated by any government entity, standards body, regulator, or other organization. The public workflow intentionally uses simplified decision-assurance logic and does not disclose proprietary algorithms, formalization methods, optimization procedures, private prompts, source code, customer information, controlled technical information, or confidential policy data. The benchmark does not establish legal or regulatory compliance, is not legal advice, and must not be interpreted as fielded, certified, independently validated, or production performance.